Go fast. In the right direction.
This is an exceptional moment. What used to take years and twenty engineers now moves very fast, when you know the platforms, the products and AI. In France and internationally.
Why now
AI spread faster than the personal computer and the internet. Three years after ChatGPT, 54.6% of Americans aged 18 to 64 were using it. At the same age, the PC had reached 19.7%.
In French companies, use is still the exception. In 2025, 18% of companies with ten or more people used AI, and 15% of those under fifty. The window is open.
It does not pay off for everyone. According to BCG, 5% of companies get value from AI at scale, and 60% get no material value. The first group posts 1.7 times more revenue growth than those 60%. The gap keeps widening in favor of those who started early.
Speed matters. Direction matters as much. In a controlled experiment with 758 participants, AI helped them finish the tasks within its reach 25% faster. On a task outside its reach, those using it found the right solution less often: 60 to 70% of the time, against 84% without AI. McKinsey sees the same thing in companies: nearly three in four of those getting the most out of AI have redesigned how they work. One in four among the rest.
My conviction: companies that move fast, in the right direction, and measure, build a lead that compounds. The others pile up experiments.
Sources: Federal Reserve Bank of St. Louis, November 2025; Insee, July 2026; BCG, September 2025; Harvard Business School, 2023; McKinsey, 2026.
What I believe
-
Speed pays. If it has a direction.
Going faster in the wrong direction is pointless. In the right one, every month ahead compounds.
-
Everywhere, and profitable.
To grow, companies first need better distribution. Being on every channel where your customers are, without losing profitability, takes more attention than one person can give. That’s where the agents come in.
-
People, with AI.
I still believe in people. An expert who masters AI, and works hand in hand with their client, goes one, two, three levels further. The agents monitor and prepare. The pilot makes the call, and answers for the result.
-
The percentage model has had its day.
I no longer believe in a percentage of media spend. I believe in sharing the value, or in days paid at their real worth: the skills to do this work are rarer than they used to be.
We didn’t add AI to the agency. We rebuilt the agency around it.
Every account has a pilot and nine agents. No agent decides in the pilot’s place.
The nine agents on an account
| Agent | What it does |
|---|---|
| The Watcher | Checks the account every morning at 7:30 a.m. |
| The Sentinel | Flags any access or change coming from outside |
| The Treasurer | Compares spend with the month’s budget |
| The Auditor | Scores the account’s management out of 100, every week |
| The Clerk | Keeps the account’s memory: every decision, dated |
| The Scout | Spots campaigns that convert but lack budget |
| The Analyst | Brings every ad platform together in a single view |
| The Reporter | Prepares the monthly report |
| The Operator | Applies a change once the pilot has approved it |
Junto’s organization from December 1, 2026.
What it changes for a brand
| Without agents | With a pilot and agents | |
|---|---|---|
| Where the budget goes | To the platforms already in place | To every channel where your customers are, measured one by one |
| Who monitors | The consultant, each time they look at the account | An agent, every morning |
| Who acts | The consultant, by hand | The agents prepare, the pilot approves, the change is applied |
| Who answers for the result | The consultant | The pilot: nothing is applied without their approval |
| Memory | In each person’s notes | In one file: every decision, dated |
| Comparison | The consultant’s experience | Industry benchmarks, aggregated and anonymized |
Who I work with
I help established brands significantly improve their media buying and their performance, and set up a new way of running it, with AI.